To enable the tracking of a plurality of objects while maintaining successiveness by including all the high intensity regions on an infrared image as objects for tracking and while detecting the presence or absence of the duplication of the objects from these duplicated data obtaining the correspondence between objects for tracking and objects for candidate.

An infrared image is A/D-transfered and binarized by an image inputting circuit 1 and a binarization circuit 2, and objects for candidate are detected by a candidate object detecting circuit 3. At the time of the first processing, a initial tracking table-generating circuit 4 registers all the objects for candidate as objects for tracking. At the time of the second processing or later, an evaluation matrix generating circuit 5 outputs the degree of presence and the degree of similarity to all the combinations between the object for candidate and the objects for tracking. A corresponding relationship obtaining circuit 6 detects a one-to-one corresponding relationship between the objects for candidate and the object for tracking. A duplication processing circuit 7 identifies a not-yet-corresponding objects for tracking as duplications on the image and detects the other duplicated objects for tracking to modify the duplicated data. An appearance determining circuit 8 registers not-yet- corresponding objects for candidate as new objects for tracking. A tracking table updating means 9 updates location predictions and characteristic data on all the objects for tracking.